of 2
Research briefing
Highly
efficient
and rapid
generation
of genetic
variants
A new mutagenesis platform
enables the fast, cost-efficient
and automatable production
of defined multi-site sequence
variants for a wide range of
applications. Demonstrations
of this method included the
generation of SARS-CoV-2 spike
gene variants, DNA fragments for
large-scale genome engineering,
and adeno-associated virus 2
(AAV2) cap genes with improved
packaging capacity.
The question
The production of kilobase-sized DNA
building blocks is a key part of design–
build–test–learn (DBTL) workflows
used for applications in biotechnol-
ogy such as synthetic biology, gene
therapy, metabolic engineering and
DNA data storage. Despite notable im-
provements in de novo DNA synthesis
in recent years
1
, current methods are
still limited by the size of DNA they can
produce and their fidelity, scale and
cost. Beyond de novo DNA synthesis,
available commercial mutagenesis kits
can produce only a limited number of
mutations and require cloning steps
that are difficult to scale. Other alterna-
tive approaches to producing genetic
variants, including recombineering,
base editing or prime editing
2
, involve
the assembly of complex constructs
and cannot be easily used in multiplex
applications. As a result of the above
shortcomings, rapid advances in the
computational design of genetic vari-
ants
3
are poised to outpace our capac-
ity to build and test DNA sequences
large enough to encode proteins in the
laboratory, thus highlighting a widen-
ing technology gap.
The discovery
DNA polymerases with proofreading
activity, such as Q5 high-fidelity DNA
polymerase, have a strong attraction to
uracil nucleotides that leads to stoppage
of DNA polymerization. However, the
modified DNA polymerase Q5U, which
contains mutations in its uracil-binding
pocket, enables the amplification of
DNA templates that contain uracil and
inosine bases. Leveraging this property,
we developed an in vitro gene variant
synthesis platform called mutagenesis
by template-guided amplicon assembly
(MEGAA). In MEGAA, a desired variant
sequence is first generated by anneal-
ing mutagenic oligonucleotides to
uracil-containing template DNA, extend-
ing these oligonucleotides with Q5U, and
ligating them with Taq DNA ligase, all in
a single-pot reaction. The process allows
6–8 nucleotide mismatches and inser-
tions, or deletions of 6–20 nucleotides,
to be easily introduced per oligonucleo
-
tide. Next, Q5 DNA polymerase is used
to amplify a desired variant amplicon
from the MEGAA reaction by PCR,
without interference from the original
uracil-containing template DNA. The re-
sulting product can then be used directly
for downstream applications.
MEGAA enables the creation of
many mutations in kilobases of DNA in
a predictable manner at an efficiency
of >90% per target site (Fig.
1a
). Itera-
tive cycling of MEGAA reactions by
using the output from one round as
the input for the next can tune the gen-
otypic purity of the desired product,
and additional combinatorial variants
if desired (Fig.
1b
). We developed an
open-source lab automation workflow
for MEGAA, dubbed MEGAAtron, to
facilitate the desktop production and
long-read sequence validation of vari-
ants. Using MEGAAtron, we demon-
strated the successful construction
of 31 natural SARS-CoV-2 spike gene
variants, 10 recoded
Escherichia coli
genome fragments each with up to 150
mutations, and 125 AAV2 gene variants
with combinatorial mutations at 6 de-
fined sites. Some of the resulting AAV2
variants exhibited tenfold enhance-
ment in viral packaging compared to
wild-type AAV2.
The implications
MEGAA is lower cost and higher
throughput for gene variant genera-
tion than traditional de novo DNA
synthesis approaches and enables the
routine production of DNA fragments
of 4–5 kilobases. This new capacity
opens up the possibility for syn-
thetic genomics projects to build and
modify genomes larger than those of
bacteria and moves the field towards
the synthesis of gigabase plant and
mammalian cell genomes engineered
to be genomically fortified and dis-
ease resistant
4
.
There are some limitations to
MEGAA. For some applications, com-
pletely pure and sequence-verified
products are needed, which would
require a clonal isolation step. Fur-
ther, the efficiency of MEGAA can be
influenced by factors such as local
G+C content, the secondary structure
of the template DNA, and the quality
of the synthesized mutagenic oligo
-
nucleotides.
Future improvements to MEGAA might
rely on orthogonally optimized DNA poly
-
merases, heat-resistant DNA ligases with
higher accuracy, desktop oligonucleotide
synthesizers, droplet-based variant synthe-
sis strategies, and more advanced predic-
tive models for DNA folding, annealing and
mutagenesis kinetics.
Liyuan Liu & Harris H. Wang
Columbia University, New York, NY, USA.
This is a summary of:
Liu, L., Huang, Y. & Wang, H. H. Fast and
efficient template-mediated synthesis of
genetic variants.
Nat. Methods
https://doi.
org/10.1038/s41592-023-01868-1
(2023)
Publisher’s note
Springer Nature remains neutral with regard
to jurisdictional claims in published maps
and institutional affiliations.
Published online: 4 May 2023
Check for updates
Nature Methods
| Volume 20 | June
2023 | 795–796
795
RefeRences
1.
Venter, J. C., Glass, J. I., Hutchison, C. A.
3rd & Vashee, S. Synthetic chromosomes,
genomes, viruses, and cells.
Cell
185
,
2708–2724 (2022).
This review highlights advances in DNA
synthesis on the horizon that will open up
new possibilities in medicine, industry,
agriculture and research.
2.
Wannier, T. M. et al. Recombineering
and MAGE.
Nat. Rev. Methods Primers
1
,
7 (2021).
This Primer article covers
recombineering, its derivative uses,
diverse applications, and methods for
combining it with other genetic editing
tools in genetic engineering.
3.
Yang, K. K., Wu, Z. & Arnold, F. H.
Machine-learning-guided directed
evolution for protein engineering.
Nat. Methods
16
, 687–694 (2019).
This article describes machine-learning
approaches to predict sequence–function
mapping without a detailed model,
thus accelerating directed evolution by
learning from characterized variants and
selecting sequences likely to improve
properties.
4.
Boeke, J. D. et al. The Genome
Project-Write.
Science
353
, 126–127 (2016).
In this article, leaders from Genome
Project-Write advocate large-scale
synthesis and engineering of gigabase
genomes of higher-order eukaryotes.
5.
Liu, L. et al. Striking antibody evasion
manifested by the Omicron variant of
SARS-CoV-2.
Nature
602
, 676–681 (2022).
This paper characterized SARS-CoV-2
Omicron variants made by on-demand
variant gene synthesis.
b
a
MEGAA
template
MEGAA
product
input
DNA
Step 1:
PCR
amplification
MEGAA
cycling
Primers
Step 2:
incorporation
p
p
p
Oligonucleotide pool
Step 3:
enrichment
by PCR
Primers
Rounds of MEGAA cycling
Design-2
Design-1
0
100
% of nanopore reads
25
50
75
1
2
3
4
5
1
3
2
9 of 9
8 of 9
7 of 9
6 of 9
5 of 9
≤ 4 of 9
Variant
types
Fig. 1 | MEGAA cycling and optimization.
a
, An overview diagram illustrating the MEGAA process.
Oligonucleotides and templates are introduced into the MEGAA workflow and assembled into variant
products that can be further cycled for improved purity.
b
, Bar plots showing the purity of gene products
generated over multiple rounds of MEGAA. Results from two mutagenic oligonucleotide design strategies
are shown: random anneal (Design-1) and ordered anneal (Design-2), highlighting improved gene product
generation through oligonucleotide optimization and MEGAA cycling. © 2023, Liu, L. et al.
expeR
t opinion
“The authors constructed this new
concept of high-throughput, in vitro
variant generation based on an existing
template, rather than de novo synthesis, to
achieve genetic variants. This concept is
fundamentally different to ‘low-throughput’
editing in vivo, such as base editing, prime
editing, the principal investor’s own previous
work MAGE (which can only introduce a
low number of variants) and the de novo
synthesis of DNA (which is expensive and
slow).”
Kaihang Wang, California Institute
of Technology, Pasadena, CA, USA.
Behind the papeR
I have been working on a synthetic
genomics project for the past few years
that involves generating sequence variants
across many DNA constructs. With our high-
throughput screening and protein structure
prediction platform in place, I needed
additional DNA constructs to functionally
test variants. Unfortunately, the COVID-19
pandemic caused substantial delays
and disruptions in the commercial DNA
synthesis sector, which greatly impacted
my work and led me to consider alternative
options. I struggled with insomnia during
this time, so I turned to regular meditation
to reduce stress. The early MEGAA
concept was conceived during a period
of meditation while I was in the shower.
It was noteworthy that we used MEGAA to
quickly generate SARS-CoV-2 spike gene
variants of the strains BA.1, BA.2 and XBB,
which improved our understanding of these
new variants of concern
5
. With invaluable
input from my mentor Harris Wang and my
co-author Yiming Huang, MEGAA became a
highly efficient and automated technology.
I owe much gratitude to my wife, Dr.
Lingling Chen, for enabling me to finish this
work.
L.L.
fR
om the editoR
“MEGAA is a method allowing the high-
efficiency construction of kilobase-sized
DNA variants, with potential applications in
the field of synthetic biology. It sidesteps the
limitations of de novo synthesis and offers
advances in multiplexability, tunability and
speed.”
Madhura Mukhopadhyay, Senior
Editor,
Nature Methods.
figuRe
Nature Methods
| Volume 20 | June
2023 | 795–796
796